Consider using pre-built Flash Attention kernels via `kernels`

未关闭
#179 2 条评论 1 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
4/5
预计耗时
3-5 天
新手友好度
35/100
Issue 类型
功能
描述清晰度
基本清楚
活跃度
冷清
技术栈
python

调研方向

Start by locating where this project uses Flash Attention and how that dependency is installed or built; the issue does not name files or tests. Compare the existing integration with the kernels package's kernels-community/flash-attn3 entry point, then determine whether supported machines and outputs meet the project's needs. Done means the maintainers agree on the integration and its scope; the issue offers a draft but does not specify acceptance criteria.

由索引模型根据 Issue 内容生成。

描述

Hey,

I am Sayak from the Kernels team at Hugging Face. I noticed that this project uses Flash Attention which includes a long build time. We ship pre-built binaries (which provide bit-exact outputs as the upstream) and thereby, we make it easy to use.

Using FA3 on a supported machine is as easy as:

# make sure `kernels` is installed: `pip install -U kernels`
from kernels import get_kernel

kernel_module = get_kernel("kernels-community/flash-attn3")
flash_attn_func = kernel_module.flash_attn_func

flash_attn_func(...)

Let us know if you'd be interested in this and and we'd be happy to provide a draft of how it would look in your repo.

主要语言
Python
星标
3.5k
派生
328
平均合并
16 小时 25 分钟
30 天内合并 PR
121

环境准备

这个项目没有提供开发容器、Dockerfile 或贡献指南,环境需要你自己搭建:先看它的 README,通用步骤见我们的新手贡献指南。

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

superlinked/sie 的其他 Issue

查看 superlinked/sie 的全部 Issue

相似的 Issue

更多 Python Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。